Home | Biodata | Biography | Photo Gallery | Publications | Tributes
Music and Musical Instruments
With the advent of high-speed general purpose digital computers, the nature of information processing in the areas of scientific, industrial and social activities has been greatly revolutionalized. During the last two decades scientists have shown keen interest in automatic pattern recognition which consists of designing a machine to mechanise a wide variety of tasks performed by humans.
The fundamental problem in pattern recognition is that of developing a recognizer and using it. For example in the automatic recognition of handwriting, the input to the recognizer is a two-dimensional curve and the output is a sequence of letters in a standard alphabet. But designing a machine to perform the task of recognizing unconstrained hand-printed characters at the level of human performance has not yet been achieved.1 However automatic recognizers have been developed for the recognition of machine-printed characters and the development has reached a stage where commercial production of the machines can be achieved. In contrast to the problem of recognition of machine-printed characters from a single font where samples of the same letter are very similar and samples of different letters are different from each other, the designer of a machine for hand-printed characters has to face the problem of defining a basis for class identity of different characters because of their wide variability in shape and style. In this article we shall describe methods developed by us for the recognition of hand-printed Tamil musical notation where the performance of the recognizer matches that of a human observer.
In recent years attempts have been made to devise a method by which the computer can recognize the musical scores directly. David Prerau of the Massachusetts Institute of Technology has evolved a method by which given a sheet of printed music in western notation, the computer recognizes the different scores and prints the names of notes with their duration in some standard notation.2 An output of the above type can be of great use for commercial music printers.
Our aim is to extend this problem to South Indian music and to devise a method by which given a sheet of printed music in Tamil musical notation, the computer recognizes the different symbols and prints them out in some standard notation. If one wants an output of the form obtained by Prerau then it will be possible to get a printout of the South Indian music in some western notation and vice versa. We have been successful in solving the first part of the problem, namely, recognizing different symbols in Tamil musical notation and the second part is yet to be attempted.
The recognition of printed South Indian music in Tamil notation is a straight forward extension of the problem of recognizing Tamil characters which we have successfully solved.3 In the Tamil musical notation we have to allow for lines drawn above or below the letters to denote the shortening of notes.
Having developed a method for printed Tamil characters, we took up the problem of recognizing hand-printed characters. With the hand-printed characters there are many different forms of the same letter and the problem becomes more complicated than the machine-printed characters.
For recognizing the different characters, we used the following methods, namely 1. Condensed run method, 2. Coded run method, 3. Coded zero-one run method developed by us. Because of the technical nature of these methods, we do not propose to describe them for the purpose of this article. However the main idea behind the application of these methods is to obtain representations in the form of strings for the different characters. (Fig. 1.)
Condensed run: 21
Condensed coded run: DC
coded run: 21
Condensed run: 1
Condensed coded run: CAC
coded run: 1G1
For the present study we have selected a song in Mohana raga from Abraham Pandithar's Karuna$mirtha Sa$gara Thirattu.4 Mohana raga makes use of five swaras namely Sa, Ri, Ga, Pa, Ta and each is represented by a letter of the Tamil alphabet. The lengthening of notes is denoted by the addition of a separate symbol. In the case of short and long forms of the swara Ri, they are denoted by two separate letters. Thus the number of major symbols to be recognized are seven letters of the Tamil alphabet.
Over 350 symbols have been chosen and digitized manually into rectangular binary matrices of zeros and ones (Fig. 1). Information is extracted in the form of strings and features that are stable in spite of the variations in shape and obtained from these strings and stored in the memory of the computer. The text which is taken up for recognition is fed into the computer and is read character by character. Each character that is read is first tested as to whether it has a horizontal line below it. If such a line exists, it is recognized and removed.The rest of the picture is reduced to a string pattern and stable features are extracted. These features are compared with the features of characters already stored in the memory. If there is agreement, the character is recognized and printed in the Roman form. We have achieved nearly cent per cent recognition rate for this problem.
Further work has to be done to improve the method of recognition of hand-printed characters. The work can be extended to the problem of recognizing printed musical scores in different fonts and to cursive handwritten musical scores.
1. R. J. Shillman (1974), Character recognition based on phenomenological attributes, theory and methods. Ph. D. Thesis, Massachusetts Institute of Technology.
2. D. Prerau (1971), Computer pattern recognition of printed music, Proceedings of the Fall Joint Computer Conference, 153-162.
3. G. Siromoney, R. Chandrasekaran and M. Chandrasekaran (1978), Computer recognition of printed Tamil characters, Pattern Recognition, 10, 243-247.
4. Abraham Pandithar (1934), Karuna$mirtha Sa$gara Thirattu, Tanjore.